人工神经网络在砂砾岩油藏采收率预测中的应用

• 作 者: 王涛 刘文锐 郭海波 朱黎明
• 机 构: 中国海油油田服务股份有限公司,中国石油吐哈油田公司勘探开发研究院,中国石化胜利油田公司东辛采油厂,中国石化中原油田公司,
• 出 处: 吐哈油气
• 发表日期: 2011年04期
• 关键词: 砾岩油藏 水驱采收率 人工神经网络 神经网络 采收率预测 开发实例 神经网络预测 开发特征 非均质 油水粘度比

目前砂砾岩油藏采收率标定存在很大难度,常规方法预测采收率与实际相比偏差很大。通过调研砂砾岩油藏开发实例及相关资料,选取我国东部某典型砂砾岩油藏,确定了砂砾岩油藏采收率的主要影响因素,并编写BP神经网络程序计算砂砾岩油藏水驱采收率,计算结果精度较高。因此,认为应用BP神经网络方法预测砂砾岩油藏水驱采收率是可行的、有效的。

Currently, it is very difficult to calibrate the recovery rate of glutenite reservoirs, and the conventional method predicts that the recovery ratio will be greatly deviated from the actual. By investigating the development of sandstone and conglomerate reservoirs and related data, a typical glutenite reservoir in eastern China was selected to determine the main factors affecting the recovery rate of glutenite reservoir. BP neural network was also used to calculate the water flooding of glutenite reservoir Yield, high accuracy of the calculation results. Therefore, it is feasible and effective to predict the water flooding recovery of glutenite reservoir by BP neural network.